PSI - Issue 53

S. Leonardi et al. / Procedia Structural Integrity 53 (2024) 327–337

332

6

S. Leonardi et al. / Structural Integrity Procedia 00 (2023) 000–000

In what follows, we measure these two types of morphological defects on the additively manufactured test samples reported in Table 4, and then investigate the role they play on the e ff ective elastic mechanical properties. Specifically, the topological defects are quantified by means of image analysis. The latter was carried out using optical micrographs of the test samples that were post-processed by means of a custom-made image analysis program developed in Python . The experimental micrographs of the sample surface were obtained via the Scanning XY function of LAS Core soft ware and a Leica DMi8 microscope. Moreover, no heat treatment was applied after fabrication and the polished test samples were then analyzed as as-built . In a nutshell, the Python program first converts the optical micrographs into binarized images, and then uses them to extract the contour of each internal pore from which the pore topological descriptors are computed. To quantify the pore aspect ratio and radius, the program uses mainly the OpenCV (4.6.0) library. Specifically, the radius is calculated from the pore area computed with the function opencv.contourArea , whereas values of the pore minor and major axes are extracted via the function opencv.fitEllipse . The elastic homogenized (i.e. e ff ective) mechanical properties of the porous materials, designed and fabricated for this study, were computed by means of FE simulations. The simulations were conducted using the commercial software Abaqus . Specifically, CPE4 elements (i.e 2D linear 4-node plane strain quadrilateral elements) and a linear elastic constitutive laws were used to model the matrix. To compute the isotropic homogenized elastic moduli, we followed the methods reported in Zerhouni et al. (2019, 2018); Tarantino et al. (2019) and specialized them for the caseof 2 D geometries under plane strain condition. Specifically, we followed the common approach of applying three linearly independent strain cases and computed the resulting average sti ff ness tensor of the porous cell (here denoting the Representative Volume Element (RVE) . To compute the average sti ff ness tensor we used a specific type of boundary conditions applied on the outer boundary of the RVE, namely the periodic boundary conditions (PBC), as given in Eq. (1) in Tarantino et al. (2019). 2.4. Finite-Element simulation framework In this section the results of the pore topological analysis conducted onto the LPBF-manufactured test samples are summarized for each metallic powder. For convenience, the reported topological descriptors are computed using the optical micrographs of the test samples topmost surface. The latter are consistent throughout the sample thickness but are not shown here for brevity. Finally, it is noted that since the focus of this work is on topological defects, the analysis of the metallurgical defects is not included hereinafter. Suxh detallurgical defects are often associated with the presence of pores and cracks within the metal matrix as documented in prior work (Zhang et al. (2019); De Terris et al. (2021); Traore et al. (2022); Du Plessis et al. (2020); Bagherifard et al. (2018)). Figures 3 and 4 summarize respectively the results of the topological analysis carried out onto the test samples produced out of AlSi10Mg and Inconel 625 powders. Notably, for each porous geometry the optical micrograph as well as the probability distributions of the pore aspect ratio and radius are reported. To statistically quantify these distributions, the mean value µ and the standard deviation σ are also given in Figures 3 and 4. Collectively, the results showthat: • heterogeneous, yet complex, internal pore features can be manufactured by LPBF with su ffi cient accuracy as revealed by comparing pairwise the experimental micrographs of the test samples (Figures 3,4) with their corresponding numerical model (Figure 2); • for all 3D-printed samples, pores that lie on the sample edge could not be accurately produced; • for both powders, the as-manufactured pores are elliptical in shape (see Figures 3d-f and 4d-f) and polydisperse in size (see Figures 3g-i and 4g-i); • for each metallic matrix, the mean value of the pore aspect ratio is similar for all cellular architectures (see Figures 3d-f and Figures 4d-f for the AlSi10Mg and the Inconel 625 test samples respectively); • irrespective of the powder employed, metallurgical defects in the form of round or irregular pores are observed within the metal matrix (see Figures 3a-c and 4a-c). 3. Results